02 / PRODUCTOne recovery loop

The system after
the alert.

Alerts are cheap. The durable workflow is knowing what matters, handing over an executable fix, and retaining the evidence that the work paid off.

LIVE NOW

Detection with an audit trail.

Sample page-level decay queue
RankPageStateLost / wkBaselineRecent
01/guides/network-cablingDECAYING4281,8421,414
02/locations/chiang-maiDECAYING237910673
03/reviews/fiber-toolsFLATLINED486481
01

Ingest

EvidenceA GSC CSV today; direct OAuth is next.

OutputComplete Monday–Sunday page histories with partial weeks removed.

02

Classify

EvidenceMedian + MAD baseline and a robust trend estimate.

OutputDECAYING, FLATLINED, HEALTHY, or INSUFFICIENT—with evidence.

03

Brief

EvidenceTop-ranked decay alerts.

OutputDiagnosis, refresh outline, and explicit recovery expectations.

04

Recover

EvidenceApplied-work date and permanent history.

OutputBefore/after evidence at four and eight weeks.

WHAT SHIPS WHEN

Honest product boundaries.

AVAILABLE CSV scan, page-level classification, ranked queue, and up to three Claude refresh briefs.

NEXT Direct GSC connection, scheduled monitoring, recovery checkpoints, and agency reports.

PLANNED Agent-readiness and business-metric modules only after the recovery loop retains customers.

Recovery signalTraffic declines, remains flat, then rises after a marked intervention.FIX APPLIED
The product becomes valuable after the intervention—not before it.